4 ms·
Strongly agree, and I'd also wish they took a shot at the Architecture market (god knows Autodesk and Archicad sorely deserves the competition) but it's not eas
by ur-whale 6y ago
Strongly agree, and I'd also wish they took a shot at the Architecture market (god knows Autodesk and Archicad sorely deserves the competition) but it's not easy to build a product that fits all possible uses cases, especially when the initial focus of the app was SFX and animation.
Specifically, for CAD, blender lacks IMO two fundamental things:
- a strong NURBS engine, complete with filleting, trimmed NURBS, CSG, full brep, etc.. (something at least on par with OpenCascade).
- a node-based (à la Houdini) modeling system to do full blown paramaterics. The current destructive + Undo/Redo model is simply not strong enough for serious CAD used.
For the latter, there are add-ons (e.g. sorcar, sverchok) that try to patch the gap, but they all feel like band-aids when you try to really put them to work.
- phkahler 6y agoAfter spending a lot of time working on the code for SolveSpace, I can say NURBS are hard. Conceptually not bad, but in practice there are a lot of complex or tricky things to do well. My hope is to one day write a new NURBS kernel in Rust. But first I need to find good solutions to those tricky problems.
- wlesieutre 6y agoI’ve recently been picking up Fusion 360 for use with a 3D printer and I’d love to be able to model this precisely with blender. I don’t love having my projects live in someone’s cloud service (for non commercial use) where they could vanish at a moment’s notice. Tried FreeCAD which is in the same vein of parametric mechanical modeling, but had a lot of trouble getting anywhere productive.